{
  "id": 352826,
  "title": "Solving DeepFake by creating another DeepFake?",
  "url": "/competitions/ai-village-ctf/discussion/352826",
  "author_name": "Vasilis Konstantakos",
  "post_date": "2022-09-15T20:31:50.446000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>So definitely not the simplest solution to the DeepFake challenge, but my idea was to create another DeepFake video to fool the detection system.</p>\n<p>I used the <a href=\"https://github.com/AliaksandrSiarohin/first-order-model\" target=\"_blank\">First Order Motion Model</a> pretrained on the VoxCeleb Dataset to generate a new (hopefully better) DeepFake.</p>\n<p>The video can be seen <a href=\"https://www.kaggle.com/code/vkonstantakos/miscellaneous-challenges-sloth-leakage-etc\" target=\"_blank\">here</a>. To create it, I just used the Google Colab notebook provided in the GitHub repository.</p>\n<p><a href=\"https://postimg.cc/7fppM413\" target=\"_blank\"><img src=\"https://i.postimg.cc/QdthDxHy/deepfake.png\" alt=\"deepfake.png\"></a></p>",
  "messages": [
    {
      "id": 1941187,
      "postDate": "2022-09-15T20:31:50.447Z",
      "content": "<p>So definitely not the simplest solution to the DeepFake challenge, but my idea was to create another DeepFake video to fool the detection system.</p>\n<p>I used the <a href=\"https://github.com/AliaksandrSiarohin/first-order-model\" target=\"_blank\">First Order Motion Model</a> pretrained on the VoxCeleb Dataset to generate a new (hopefully better) DeepFake.</p>\n<p>The video can be seen <a href=\"https://www.kaggle.com/code/vkonstantakos/miscellaneous-challenges-sloth-leakage-etc\" target=\"_blank\">here</a>. To create it, I just used the Google Colab notebook provided in the GitHub repository.</p>\n<p><a href=\"https://postimg.cc/7fppM413\" target=\"_blank\"><img src=\"https://i.postimg.cc/QdthDxHy/deepfake.png\" alt=\"deepfake.png\"></a></p>",
      "rawMarkdown": "So definitely not the simplest solution to the DeepFake challenge, but my idea was to create another DeepFake video to fool the detection system.\n\nI used the [First Order Motion Model](https://github.com/AliaksandrSiarohin/first-order-model) pretrained on the VoxCeleb Dataset to generate a new (hopefully better) DeepFake.\n\nThe video can be seen [here](https://www.kaggle.com/code/vkonstantakos/miscellaneous-challenges-sloth-leakage-etc). To create it, I just used the Google Colab notebook provided in the GitHub repository.\n\n[![deepfake.png](https://i.postimg.cc/QdthDxHy/deepfake.png)](https://postimg.cc/7fppM413)",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1941187": "So definitely not the simplest solution to the DeepFake challenge, but my idea was to create another DeepFake video to fool the detection system.\n\nI used the [First Order Motion Model](https://github.com/AliaksandrSiarohin/first-order-model) pretrained on the VoxCeleb Dataset to generate a new (hopefully better) DeepFake.\n\nThe video can be seen [here](https://www.kaggle.com/code/vkonstantakos/miscellaneous-challenges-sloth-leakage-etc). To create it, I just used the Google Colab notebook provided in the GitHub repository.\n\n[![deepfake.png](https://i.postimg.cc/QdthDxHy/deepfake.png)](https://postimg.cc/7fppM413)"
  }
}